KernelProcessor<T> Class

public abstract class KernelProcessor<T>

Public MustInherit Class KernelProcessor(Of T)

generic<typename T>
public ref class KernelProcessor abstract

[<AbstractClassAttribute>]
type KernelProcessor<'T> = class end

Inheritance

Object  →  KernelProcessor<T>

Derived

Numerics.NET.SignalProcessing.ComplexKernelProcessor<T>

Numerics.NET.SignalProcessing.RealKernelProcessor<T>

Type Parameters

T

The element type (T for real kernels, Complex<T> for complex kernels).

This class provides high-performance 1D convolution and correlation operations with flexible boundary handling and output modes. It supports both real and complex-valued signals.

Convolution vs. Correlation:

• Convolution: y[n] = Σ_m x[n - (m - a)] · h[m], where a is the anchor index. This is the standard operation for filtering and feature detection.
• Correlation: y[n] = Σ_m x[n - (m - a)] · conj(h[m]). This is used for template matching and signal similarity measurement.

Comparison with other libraries:

• NumPy (np.convolve): Equivalent to using FullLength, SameAsSignal, or NoPadding with Default. Only supports Zero.
• SciPy (scipy.signal.convolve): Similar to NumPy but with additional boundary modes. The mode parameter maps to ConvolutionMode, and method maps to KernelProcessingMethod. Use Origin(Int32) for scipy-style origin specification.
• OpenCV (cv2.filter2D): Typically uses centered kernels with replicate padding. Equivalent to Centered with Replicate.

Reference